Objective: The optimal management of Barrett's esophagus (BE) is controversial. Little is known about current practice patterns or associated direct medical costs.
Methods: In a national cross-sectional survey, we asked a random sample of gastroenterologists how they would manage patients with BE and various degrees of dysplasia. We used logistic regression to identify factors associated with so-called "frequent" (at least every 12 months) surveillance. We calculated direct medical costs using Medicare payments and population-based estimates of the number of BE patients under surveillance.
Results: Approximately 50% of 555 gastroenterologists responded. More than 96% of respondents recommended endoscopic surveillance for BE. For BE without dysplasia, 30% would perform frequent surveillance; this was the case particularly gastroenterologists older than age 45 yr (odds ratio = 1.91, p = 0.038) or those receiving primarily fee-for-service reimbursement (odds ratio = 2.57, p = 0.004). For BE with low-grade dysplasia, the frequency of endoscopy was highly variable (range, 1-24 months). For BE with high-grade dysplasia, 73% of gastroenterologists recommended esophagectomy and the remainder recommended endoscopic surveillance. Approximately 95% of the gastroenterologists who recommended surveillance for high-grade dysplasia, however, were not in agreement with recommended protocols. We estimated the national annual expenditure for surveillance endoscopy every 24 months for BE without dysplasia to be at least $22 million. Increase in surveillance intensity from low frequency (every 36 months) to high frequency (every 12 months) strategies would escalate costs by $29 million annually.
Conclusions: Physician age and reimbursement influence BE surveillance practice, suggesting the influence of nonclinical factors on clinical decision making. The majority of clinicians who would recommend surveillance for high-grade dysplasia may not be using an appropriately aggressive strategy. Variations in surveillance strategies can have large cost implications.